Inter-individual susceptibility to mutagens/carcinogens can be assessed by either genotyping DNA repair genes in different pathways or phenotyping DNA repair capacity (DRC) at the molecular or cellular level. Due to the large number of known DNA repair genes, and the interactions between repair pathways, phenotyping is becoming the preferred approach to measure DRC, and reliable assays are therefore increasingly needed. The use of a cellular phenotype comet assay for the nucleotide excision repair (NER) pathway using benzo[a]pyrene diol epoxide (BPDE) has been described in previous papers, but no thorough evaluation of its applicability in large genotype-phenotype studies has been presented. Our aim was to evaluate the possibility of using cryopreserved instead of fresh peripheral blood mononuclear cells (PBMCs) to evaluate intra- and inter-assay variation, and inter-individual variation, for the aphidicolin (APC)-block NER comet assay. Moreover, we measured the variation for the designated internal standard (K562 erythroleukaemia cell line) and we evaluated the feasibility to use lymphoblastoid cell lines (LCLs) as surrogate of PBMCs. Our results showed a low intra-assay [coefficient of variation (CV) 19.9%] and inter-assay (CV 32.3%) variation, with a good inter-individual variation (122 subjects, mean ± standard deviation 7.38 ± 4.99; range 0.66-26.14; CV 67.63%). A significant correlation between results derived from cryopreserved and fresh PBMCs from the same individuals was found (10 subjects, r = 0.62, P = 0.05). Results from LCLs and cryopreserved PBMCs from the same subjects showed an inverse significant correlation (10 subjects, r = -0.712, P = 0.02). K562 cells as internal standard showed low intra-assay variation. In the present study the APC-block NER comet assay on cryopreserved PBMCs seemed to be a reliable method to measure DRC variation in epidemiological studies; LCLs were not a good surrogate in this assay.